% Bahn analysis report % author: “Jinshi” % date: “March 26, 2019”
| output: |
| html_document: |
| df_print: paged |
| bibliography: bibliography.ris |
1. The spatial destribution of global Rs sites
- We have much more measured Rs from the middle latitude region
- It is difficult to measure soil respiration all year around in the cold region
- Developed contries most located at Middle latitude region, thus recieved more funds to support the Rs measurements
Global spatial destribution of soil repiration sites
It is obvious that Rs measurements from cold region is more importent, but how? Improve the Rs measure equipment so it can measure Rs in cold condition; Increasing funds; * Measured once per day to get daily mean 
Rs measured at diurnal soil temperature
- Measure once per year to get a annual Rs mean? Bahn’s approach [Bahn et al. (2004) Biogeosciences]
Rs measured at mean annual soil temperature
- Rs measured at annuam mean temperature linearly related with annual Rs rate
- Rs at mean temperature: soil respiration measured at annual mean temperature / monthly mean temperature / daily mean temperature
- Rs_annual ~ Rs_mat (Rs_bahn)
- Rs_annual = 436.2 * Rs_mat^0.926 (R^2 = 0.94, p<0.001)
2. The object of this analysis are
- Whether Rs measured at annuam mean temperature can represent annual Rs rate?
- If not, when and what is the mechanism?
- Update the model?
3. Mthods
- Data
- Statistic analysis
- Modeling
3.1 Data used from SRDB_V4
- Climate space figure
## Warning: Removed 3 rows containing missing values (geom_point).## Saving 7 x 5 in image## Warning: Removed 3 rows containing missing values (geom_point).
3.2 Ts sources (MGRsD, MGRsD_TAIR, From paper, Rs_Ts_relationship)
## Saving 7 x 5 in image
3.3 Annual Rs or Ts coverage effect
## Saving 7 x 5 in image
## Saving 7 x 5 in image
3.4.1 Effect of maximum allowed divergence between global climate data set and site-specific air temperature
- Does TAIR_dev and TAIR_LT<_dev affect the relationship – YES!!!!!
- Does TAIR_LTM_dev () pull the slope off 1? – YES!!!!!
- TAIR_LTM_dev = with( srdb, abs( MAT_Del - MAT ) )
- TAIR_dev <- with( srdb, abs( TAnnual_Del - Study_temp ) )
- Figure E. Effect of maximum allowed divergence between global climate data set and site-specific air temperature, when given. As we throw out data points with high divergence, R2 goes up (top panel) and RSE goes down (bottom, g C m-2 yr-1).
- This is different from what SPI test
## Warning: Removed 302 rows containing missing values (geom_errorbarh).## Saving 7 x 5 in image## Warning: Removed 302 rows containing missing values (geom_errorbarh).
## Saving 7 x 5 in image
3.4.2 Effect of maximum allowed divergence between annual precipitation from paper and Del
3.5 test the relationship between Rs_annual and Rs_mat
## Fri Mar 29 18:02:52 2019 -------------------+++++-------------------
## Fri Mar 29 18:02:52 2019 Bahn relationship for these data:
##
## Call:
## lm(formula = Rs_annual ~ Rs_TAIR, data = sdata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -746.75 -111.49 -39.62 86.43 1273.96
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 160.04546 13.25471 12.07 <2e-16 ***
## Rs_TAIR 0.94510 0.01624 58.21 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 194.7 on 821 degrees of freedom
## Multiple R-squared: 0.8049, Adjusted R-squared: 0.8047
## F-statistic: 3388 on 1 and 821 DF, p-value: < 2.2e-16
4. Results
4.1 Using Ts, TAnnual or MAT
4.1.1 Using soil temperature
##
## Call:
## lm(formula = bahn ~ Rs_annual, data = sdata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1269.9 -106.0 18.3 117.3 1222.3
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -4.58745 17.22279 -0.266 0.79
## Rs_annual 1.07455 0.01846 58.222 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 233.1 on 821 degrees of freedom
## Multiple R-squared: 0.805, Adjusted R-squared: 0.8048
## F-statistic: 3390 on 1 and 821 DF, p-value: < 2.2e-16
## Fri Mar 29 18:02:53 2019 -------------------+++++-------------------
## Fri Mar 29 18:02:53 2019 How are Rs_annual and Rs_annual_bahn_Temp related?
## Fri Mar 29 18:02:53 2019 sdata rows = 823 cols = 142
## Fri Mar 29 18:02:53 2019 Model summary:
##
## Call:
## lm(formula = temp ~ Rs_annual, data = sdata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1269.9 -106.0 18.3 117.3 1222.3
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -4.58745 17.22279 -0.266 0.79
## Rs_annual 1.07455 0.01846 58.222 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 233.1 on 821 degrees of freedom
## Multiple R-squared: 0.805, Adjusted R-squared: 0.8048
## F-statistic: 3390 on 1 and 821 DF, p-value: < 2.2e-16
##
## Fri Mar 29 18:02:53 2019 Plotting and saving model diagnostics...## Fri Mar 29 18:02:53 2019 Plotting and saving model residuals...## Fri Mar 29 18:02:53 2019 Saving outputs/3-modelresids.pdf## Saving 4 x 2.8 in image
## Fri Mar 29 18:02:53 2019 Test H0 of intercept=0: p-value = 0.7900293
## Fri Mar 29 18:02:53 2019 Test H0 of slope=1: p-value = 5.863901e-05## [1] 0.7900293## [1] 5.863901e-054.1.2 Using T_Annual
##
## Call:
## lm(formula = bahn ~ Rs_annual, data = sdata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2450.0 -173.0 -0.3 151.8 4964.6
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -29.30007 27.53434 -1.064 0.288
## Rs_annual 1.03037 0.02951 34.921 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 372.7 on 821 degrees of freedom
## Multiple R-squared: 0.5976, Adjusted R-squared: 0.5971
## F-statistic: 1219 on 1 and 821 DF, p-value: < 2.2e-16
## Fri Mar 29 18:02:54 2019 -------------------+++++-------------------
## Fri Mar 29 18:02:54 2019 How are Rs_annual and Rs_annual_bahn_Temp related?
## Fri Mar 29 18:02:54 2019 sdata rows = 823 cols = 142
## Fri Mar 29 18:02:54 2019 Model summary:
##
## Call:
## lm(formula = temp ~ Rs_annual, data = sdata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2450.0 -173.0 -0.3 151.8 4964.6
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -29.30007 27.53434 -1.064 0.288
## Rs_annual 1.03037 0.02951 34.921 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 372.7 on 821 degrees of freedom
## Multiple R-squared: 0.5976, Adjusted R-squared: 0.5971
## F-statistic: 1219 on 1 and 821 DF, p-value: < 2.2e-16
##
## Fri Mar 29 18:02:54 2019 Plotting and saving model diagnostics...## Fri Mar 29 18:02:54 2019 Plotting and saving model residuals...## Warning: Removed 2 rows containing non-finite values (stat_smooth).## Warning: Removed 2 rows containing missing values (geom_point).## Fri Mar 29 18:02:54 2019 Saving outputs/3-modelresids.pdf## Saving 4 x 2.8 in image## Warning: Removed 2 rows containing non-finite values (stat_smooth).
## Warning: Removed 2 rows containing missing values (geom_point).
## Fri Mar 29 18:02:54 2019 Test H0 of intercept=0: p-value = 0.2875834
## Fri Mar 29 18:02:54 2019 Test H0 of slope=1: p-value = 0.3037032## [1] 0.2875834## [1] 0.30370324.1.3 Using MAT
##
## Call:
## lm(formula = bahn ~ Rs_annual, data = sdata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2356.27 -167.25 -1.34 142.82 2746.39
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -32.36603 24.27741 -1.333 0.183
## Rs_annual 1.00324 0.02602 38.563 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 328.6 on 821 degrees of freedom
## Multiple R-squared: 0.6443, Adjusted R-squared: 0.6439
## F-statistic: 1487 on 1 and 821 DF, p-value: < 2.2e-16
## Fri Mar 29 18:02:55 2019 -------------------+++++-------------------
## Fri Mar 29 18:02:55 2019 How are Rs_annual and Rs_annual_bahn_Temp related?
## Fri Mar 29 18:02:55 2019 sdata rows = 823 cols = 142
## Fri Mar 29 18:02:55 2019 Model summary:
##
## Call:
## lm(formula = temp ~ Rs_annual, data = sdata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2356.27 -167.25 -1.34 142.82 2746.39
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -32.36603 24.27741 -1.333 0.183
## Rs_annual 1.00324 0.02602 38.563 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 328.6 on 821 degrees of freedom
## Multiple R-squared: 0.6443, Adjusted R-squared: 0.6439
## F-statistic: 1487 on 1 and 821 DF, p-value: < 2.2e-16
##
## Fri Mar 29 18:02:55 2019 Plotting and saving model diagnostics...## Fri Mar 29 18:02:55 2019 Plotting and saving model residuals...## Warning: Removed 1 rows containing non-finite values (stat_smooth).## Warning: Removed 1 rows containing missing values (geom_point).## Fri Mar 29 18:02:55 2019 Saving outputs/3-modelresids.pdf## Saving 4 x 2.8 in image## Warning: Removed 1 rows containing non-finite values (stat_smooth).
## Warning: Removed 1 rows containing missing values (geom_point).
## Fri Mar 29 18:02:55 2019 Test H0 of intercept=0: p-value = 0.1828444
## Fri Mar 29 18:02:55 2019 Test H0 of slope=1: p-value = 0.9008364## [1] 0.1828444## [1] 0.90083644.2 Analysis when Rs_mat cannot represent Rs_annual
4.2.1 Test extreme high Rs values (>3000)
- unlikely these extremes high Rs values have big influence
## Warning: Removed 510 rows containing missing values (geom_errorbarh).## Saving 4 x 2.8 in image## Warning: Removed 510 rows containing missing values (geom_errorbarh).
## Saving 4 x 2.8 in image
4.2.2 Does Ecosystem_type affects the relationship between Rs_annual and Rs_mat
## Warning: Removed 495 rows containing missing values (geom_errorbarh).## Saving 7 x 5 in image## Warning: Removed 495 rows containing missing values (geom_errorbarh).
## Saving 7 x 5 in image
## Warning: Removed 16 rows containing missing values (geom_errorbarh).## Saving 7 x 5 in image## Warning: Removed 16 rows containing missing values (geom_errorbarh).
## Saving 7 x 5 in image
## Warning: Removed 503 rows containing missing values (geom_errorbarh).## Saving 7 x 5 in image## Warning: Removed 503 rows containing missing values (geom_errorbarh).
## Saving 7 x 5 in image
## Warning: Removed 8 rows containing missing values (geom_errorbarh).## Saving 7 x 5 in image## Warning: Removed 8 rows containing missing values (geom_errorbarh).
## Saving 7 x 5 in image
4.2.3 Does Meas_method affects the relationship
## Warning: Removed 54 rows containing missing values (geom_errorbarh).## Saving 7 x 5 in image## Warning: Removed 54 rows containing missing values (geom_errorbarh).
## Saving 7 x 5 in image
## Warning: Removed 457 rows containing missing values (geom_errorbarh).## Saving 7 x 5 in image## Warning: Removed 457 rows containing missing values (geom_errorbarh).
## Saving 7 x 5 in image
4.2.4 RA or RH dominated sites differ?
## Saving 7 x 5 in image
4.2.5 Biome effect
4.2.6 TAIR and precipitation variability affect?
## Saving 7 x 5 in image
## Saving 7 x 5 in image
4.2.7 Does drought affect?
## Saving 7 x 5 in image
## Warning in qt((1 - level)/2, df): NaNs produced
5. Monthly and daily results
6. More analysis in the future
- 1 T&Drought function (Maybe use PDSI)
- 2 seprete out Agriculture & Wetland
- 3 Using SD information with boosting?
- 4 Use Rs_mat predict Rh?
- 5 Use this approach estimate global Rs
- 6 Think about application